Converting a string to code

Noteworthy points:

• I'm new to coding and am testing various things to learn;
• i.e. yes, I'm sure there are better ways do achieve what I am trying to do;
• I would like to know any alternative / more efficient methods, however;
• I would also still like to know how to convert string to code to achieve my goal with this technique

So far I have looked around the forum and on google, and seen a few topics on this, none of which I can made work here, or which precisely answer the question from my perspective, including using eval and exec.

The Scenario

• I have a dataframe: london with 23 columns
• I want to create a dataframe showing all rows with 'NaN' values
• I have tried to use .isnull(), but it appears to only work on a single column at a time
• I am trying to achieve my desired result by using | to return any rows in any columns where .isnull() returns True

An example of this working with just two columns is:

london[(london['Events'].isnull() | london['Max Gust SpeedKm/h'].isnull())]

However, I need to achieve this result with all 23 columns, so I have attempted to complete this with some code.

Attempted Solution

• Creating a string containing all of the column headers
• i.e. london[(london['Column Header'].isnull() followed by | and then the next column
• Then using this string within the container shown in the working example above
• i.e. london[(string)]

I have managed to create the string I need using the following:

string = []
for i in (london.columns.values):
    string.append("london['" + i + "'].isnull()")
    string.append(" | ")
del string[-1]
final_string = "".join(string)

And finally when I try to implement the final step, I cannot work out how to convert this string into usable code.

For example:

now = eval(final_string)
london[now]

Resulting in:
NotImplementedError: 'Call' nodes are not implemented

Thank you in advance.